Oral steroids for adhesive capsulitis

Abstract

Background

This review is one in a series of Cochrane reviews of interventions for shoulder pain in adults.

Objectives

To determine the efficacy and safety of oral steroids for adhesive capsulitis.

Search methods

Searches of the Cochrane Library including CENTRAL, Issue 4, 2005, Cochrane Musculoskeletal Review Group Register, MEDLINE, EMBASE, CINAHL were conducted in November 2005, unrestricted by date or language.

Selection criteria

Only studies described as randomised controlled trials studying participants with adhesive capsulitis, frozen shoulder, stiff painful shoulder or periarthritis and interventions of oral steroids compared to placebo, no treatment, or any other treatment were included.

Data collection and analysis

Two independent reviewers assessed methodological quality of each included trial and extracted data. Standard Cochrane methodology was used to analyse the extracted data.

Main results

Five small trials were included: two trials (30 and 49 participants) of oral steroids or placebo; one trial (40 participants) of oral steroids or no treatment; one trial (28 participants) of oral or intra‐articular steroids; and one trial (32 participants) of manipulation under anaesthesia and intraarticular steroid injection with or without oral steroids. Study participants were similar across trials, but no trial used the same oral steroid regimen or dosage. Trials were of variable quality (only one of high quality) and some were poorly reported.

No meta‐analyses could be performed as no raw data could be extracted from one placebo‐controlled trial and three trials used different comparators. One trial reported significant short‐term benefits of oral steroids versus placebo: 48% more participants reported success (RR = 2 (95% CI 1.3 to 3.1, NNT=2); overall improvement in pain 2.7 (95% CI 1.4 to 4.0) on a 0 to 10 point scale; total shoulder abduction increased by 23.3 degrees (95% CI 11.3 to 35.3); Shoulder Pain and Disability Index (SPADI) score improved by 18.1 (95% CI 7.6 to 28.6) on a 0 to 100 point scale. But benefits were not maintained at 6 weeks. A second trial reported no significant differences between oral steroid and placebo in pain or range of movement but it suggested improvement occurred earlier in the steroid treated group. A third trial reported that oral steroids provided a more rapid initial improvement in pain compared to no treatment but negligible differences by five months. There were minimal adverse effects reported.

Authors' conclusions

Available data from two placebo‐controlled trials and one no‐treatment controlled trial provides "Silver" level evidence (www.cochranemsk.org) that oral steroids provides significant short‐term benefits in pain, range of movement of the shoulder and function in adhesive capsulitis but the effect may not be maintained beyond six weeks.

Plain language summary

Oral steroids for shoulder pain (adhesive capsulitis)

This summary of a Cochrane review presents what we know from research about the effect of steroids taken as pills (oral) for adhesive capsulitis. The review shows that:

There is silver level evidence (www.cochranemsk.org) that oral steroids may work to treat shoulder pain (adhesive capsulitis) in the short term. Oral steroids may decrease pain and disability, and may improve movement in the shoulder in the short term. But the benefits of oral steroids may not last 6 weeks. Oral steroids taken for short periods in people who are otherwise healthy may not cause harms. There is not enough evidence to be certain of the benefits and harms of oral steroids and more research is needed.

What is adhesive capsulitis and what drugs are used to treat it? 
 Shoulder pain can be caused by a number of different conditions. It can be caused by rotator cuff disease or adhesive capsulitis (also called frozen shoulder, stiff painful shoulder or periarthritis). While both conditions are painful, adhesive capsulitis also tends to cause stiffness in the shoulder no matter which way you move it. The pain and stiffness in the shoulder can go away on its own but could last up to 2 to 3 years. Some people may still not be able to move their shoulder fully after 3 years.

Drug and non‐drug treatments are used to relieve the pain and stiffness. In other arthritis diseases, steroids, taken as pills, have been shown to work. It is therefore thought that steroids, such as prednisolone or cortisone pills, may work for adhesive capsulitis.

What are the results of this review? 
 The studies tested people who had adhesive capsulitis for about 6 months. They were given no treatment, fake treatments, steroid injections or oral steroids. Oral steroids, such as prednisolone or cortisone were given for about 3 to 4 weeks, and sometimes again for another 3 to 4 weeks if people still had pain and stiffness. All people had physiotherapy or an exercise programme while taking the steroids.

Benefits of oral steroids 
 In people with adhesive capsulitis, at 3 weeks, oral steroids

may work more than fake pills

­48 out of 100 people who took fake pills said they were better 
 ­96 out of 100 people who took steroids said they were better

may decrease pain and disability more than fake pills

­pain may decrease by 2.7 more points on a scale of 0 to 10 with steroids 
 ­disability may decrease by 18 more points on a scale of 0 to 100 with steroids

may increase the ability to move the shoulder more than fake pills

­shoulder movement increased by 23 degrees 
 But these benefits did not last as long as 6 weeks so there is not enough evidence to be certain of the results beyond 3 weeks.

Oral steroids may also improve pain earlier and quicker than no treatment at all. But after 5 months there were no benefits of oral steroids over no treatment. There is also not enough evidence to be certain of the results.

Harms of oral steroids 
 In people with adhesive capsulitis who have no serious other problems, taking oral steroids for a short time may not cause serious side effects. But there is not enough evidence to be certain. Other research about steroids taken over longer periods of time shows that harms could include high cholesterol and high blood pressure.

Background

This Cochrane review is one of a series of Cochrane reviews of interventions for shoulder disorders. This series of reviews form the update of an earlier Cochrane Review of all interventions for shoulder disorders (Green 1998a; Green 1998b). In our previous systematic review, we concluded that there was not enough evidence to either support or refute the efficacy of any of the commonly used interventions for this condition including non‐steroidal anti‐inflammatory drugs, corticosteroid injections, oral steroids and physiotherapy, and further well‐designed clinical trials were needed (Green 1998a, Green 1998b). Since this time many new clinical trials, studying a diverse range of interventions, have been performed. Thus, for updating, we subdivided the review into a series of reviews investigating the evidence for efficacy of single interventions, and where applicable, for specific shoulder disorders. This review examines the evidence for efficacy and safety of oral steroids for the treatment of adhesive capsulitis.

Adhesive capsulitis (also termed frozen shoulder, stiff painful shoulder or periarthritis) is a common cause of shoulder pain estimated to affect 2‐5% of the general population (Anton 1993, Lundberg 1969). The cumulative incidence of presentations to general practice from a Dutch study of shoulder complaints has been estimated to be 2.4/1000/year (95% CI 1.9‐2.9) (van der Windt 1995). Duplay first described a condition of painful stiffening of the shoulder in 1872 (Duplay 1872). Since then other terms have been used to label this condition including 'frozen shoulder', to describe painful restriction of range of motion of the shoulder with normal plain roentgenographs (Codman 1934) and 'adhesive capsulitis', based upon the shoulder joint arthrographic findings (Neviaser 1945).

The disorder is characterized by spontaneous onset of shoulder pain and progressive global stiffness of the glenohumeral joint accompanied by significant disability (Reeves 1975, Neviaser 1987). Most studies have suggested a self‐limiting condition lasting an average of two to three years although significant numbers of people have residual clinically detectable restriction of movement beyond three years and smaller numbers have residual disability (Reeves 1975, Grey 1978, Hazleman 1972, Clarke 1975, Binder 1984, Lloyd‐Roberts 1959, Simmonds 1949).

Several interventions are advocated to treat adhesive capsulitis, although we have previously found limited data from randomised controlled trials to support their use (Green 1998a, Green 1998b). Oral steroids were first used in the 1950's; anecdotal evidence suggests expedited recovery and reduced need for manipulation under anaesthesia with oral steroid use (Coventry 1953, Sigler 1951, Ehrlich 1951) while more recent evidence regarding efficacy is equivocal. However, as there is good evidence that oral steroids such as prednisolone and prednisone dramatically reduce the symptoms of inflammatory joint disease such as rheumatoid arthritis (Gotzsche 1998), oral steroids may also be effective in adhesive capsulitis, particularly early in the course of the disease.

Objectives

To determine the efficacy and safety of oral steroids for adhesive capsulitis.

Methods

Criteria for considering studies for this review

Types of studies

Only studies described as randomised controlled trials (RCTs) were included in this systematic review.

Types of participants

Participants described as having adhesive capsulitis, frozen shoulder, stiff painful shoulder or periarthritis were eligible for inclusion. Studies that included mixed populations of participants with shoulder pain were only included provided that results for the adhesive capsulitis participants were presented separately or > 90% of participants in the study had adhesive capsulitis.

Types of interventions

All randomised controlled trials of oral steroids compared to placebo, no treatment, or any other treatment; or randomised controlled trials comparing varying types and dosages of oral steroids were included.

Types of outcome measures

All outcomes measured in the trials were reported. These included pain (overall, at night, at rest, and on movement), range of motion (active and/or passive: flexion, abduction, external rotation, internal rotation and hand behind back), function, quality of life, treatment success and adverse effects.

Search methods for identification of studies

We searched the following electronic databases and sources to identify studies: 
 1. MEDLINE (OVID) from 1966 to November 2005; 
 2. EMBASE (OVID) from 1966 to November 2005; 
 3. CINAHL (OVID) from 1982 to November 2005; 
 4. Science Citation Index (SCISEARCH) from 1966 to November 2005; 
 5. Cochrane Library, including Cochrane Central Register of Controlled Trials (CENTRAL) Issue 4, 2005; 
 6. The Cochrane Musculoskeletal Review Group Registry; 
 7. Reference lists in review articles and trials retrieved; 
 8. Personal communication with experts in the field.

We combined keywords/textwords describing adhesive capsulitis and terms describing the intervention to identify randomised controlled trials. There were no language or date restrictions. We conducted all searches on November 29‐30, 2005. The MEDLINE search strategy (Appendix 1) combined the Cochrane Highly Sensitive Search Strategy, phases one and two, as described in the Cochrane Handbook for Systematic Reviews of Interventions 4.2.5 (Higgins 2005) with search terms describing the condition and the intervention.

We adapted the search strategy for the CENTRAL, EMBASE and CINAHL databases as appropriate; these search strategies are outlined in Additional Tables, Appendix 2.

Data collection and analysis

STUDY SELECTION 
 For this updated review, we generated the electronic searches in MEDLINE, EMBASE, CINAHL and CENTRAL and downloaded the citations into Endnote 7.0. Two of the authors (RB, JY, SG or RJ) then independently reviewed the information to identify trials that could potentially meet the inclusion criteria. Full articles describing these trials were obtained and the same two authors independently applied the selection criteria to the studies. There was complete consensus concerning the final inclusion of RCTs.

METHODOLOGICAL QUALITY ASSESSMENT 
 Two reviewers (not always the same pair) independently assessed the methodological quality of each RCT. Authors were not blinded with respect to authors, institution and journal because they were familiar with the literature. We used consensus to resolve disagreements and a third reviewer if disagreements persisted.

As in the previous review, we assessed methodological quality based upon whether the trials met key criteria (appropriate randomisation, allocation concealment, blinding, a priori calculation of sample size, number lost to follow up and intention to treat analysis). Failure to fulfil these criteria were considered to have potentially biased the overall outcome of the included trial. Allocation concealment was ranked as: A: adequate; B: unclear; C: inadequate; or D: not used. All other information concerning the above criteria was recorded on a pre‐piloted data extraction sheet and later transposed into the Table of Included Studies.

DATA EXTRACTION 
 Two reviewers independently extracted study characteristics data including source of funding, study population, intervention, analyses and outcomes using standardised data extraction forms. The authors of recent original studies were contacted to obtain more information when needed. Those that were contacted has been recorded in the notes section of the Table of Characteristics of included studies.

In order to assess efficacy, raw data for outcomes of interest (means and standard deviations for continuous outcomes and number of events for dichotomous outcomes) were extracted where available in the published reports. Wherever reported data was converted or imputed, this was recorded in the notes section of the Table of Characteristics of included studies.

ANALYSIS 
 The results of each RCT were plotted as point estimates, i.e., relative risks with corresponding 95% confidence interval for dichotomous outcomes, and mean and standard deviation for continuous outcomes. To expedite rapid and easier updating of the review we report all results that could be extracted from the included trials. When the results could not be shown in this way, they were described in the Table of Characteristics of Included Studies. For continuous measures, preference was given to analyse the results with weighted mean differences because these results are easier to interpret for clinicians/readers. The studies were first assessed for clinical homogeneity with respect to the duration of the disorder, control group and outcomes. For studies judged to be clinically heterogeneous we planned to describe them separately and not combine them in a meta‐analysis. For studies judged as clinically homogeneous, we planned to test statistical heterogeneity using Q test (chi‐square) and I². We planned to pool clinically and statistically homogeneous studies using the fixed effects model, and clinically homogeneous and statistically heterogeneous studies using the random effects model. A sensitivity analysis was planned to assess for any bias attributable to allocation concealment.

CLINICAL RELEVANCE TABLES 
 Clinical relevance tables were compiled under additional tables for selected important, and statistically significant outcomes, to improve the readability of the review. In the clinical relevance tables, for dichotomous outcomes (e.g., patient reported success), the baseline risk was entered directly from the observed events in the control group displayed on the RevMan Metaview screen. The control (placebo) event rate (expressed as a percentage) was used. It is the sum of all the events in the placebo group divided by the total patient numbers in the placebo group. The NNT was calculated as one divided by the absolute risk difference, for outcomes derived from one trial. For continuous outcomes (e.g., overall pain measured on a visual analogue scale), absolute change (benefit) was calculated from the weighted mean difference and expressed as a percent and in the original units. Relative difference in the change from baseline was calculated as the absolute benefit divided by the baseline mean of the control (placebo) group.

GRADING THE STRENGTH OF THE EVIDENCE 
 The common system of grading the strength of scientific evidence for a therapeutic agent that is described in the CMSG module scope and in the Evidence‐based Rheumatology BMJ book (Tugwell 2003) and was used to rank the evidence included in this systematic review. Four categories are used to rank the evidence from research studies from highest to lowest quality: Platinum, Gold, Silver, and Bronze.

Platinum: A published systematic review that has at least two individual controlled trials each satisfying the following : 
 ·Sample sizes of at least 50 per group ‐ if these do not find a statistically significant difference, they are adequately powered for a 20% relative difference in the relevant outcome. 
 ·Blinding of patients and assessors for outcomes. 
 ·Handling of withdrawals >80% follow up (imputations based on methods such as Last Observation Carried Forward (LOCF) are acceptable). 
 ·Concealment of treatment allocation.

Gold: At least one randomised clinical trial meeting all of the following criteria for the major outcome(s) as reported: 
 ·Sample sizes of at least 50 per group ‐ if these do not find a statistically significant difference, they are adequately powered for a 20% relative difference in the relevant outcome. 
 ·Blinding of patients and assessors for outcomes. 
 ·Handling of withdrawals > 80% follow up (imputations based on methods such as LOCF are acceptable). 
 ·Concealment of treatment allocation.

Silver: A systematic review or randomised trial that does not meet the above criteria. Silver ranking would also include evidence from at least one study of non‐randomised cohorts that did and did not receive the therapy, or evidence from at least one high quality case‐control study. A randomised trial with a 'head‐to‐head' comparison of agents would be considered silver level ranking unless a reference were provided to a comparison of one of the agents to placebo showing at least a 20% relative difference.

Bronze: The bronze ranking is given to evidence if at least one high quality case series without controls (including simple before/after studies in which patients act as their own control) or if the conclusion is derived from expert opinion based on clinical experience without reference to any of the foregoing (for example, argument from physiology, bench research or first principles).

In this review, as only RCTs were included, the bronze ranking of evidence was not applicable. The ranking is included in the synopsis and abstract of this review, and in the clinical relevance tables (in Additional Tables).

Results

Description of studies

Seven potential trials from 443 citation postings were identified and five small trials including 40, 30, 49, 32 and 28 participants respectively met the inclusion criteria (Binder 1986; Blockey 1954; Buchbinder 2004; Kessel 1981; Widiastuti‐Samekto). Details of the five included trials are given in the Table of Characteristics of Included Studies. All trials were published in English.

Two trials were excluded because they were non‐randomised comparisons (Lloyd‐Roberts 1959; Melzer 1995).

INTERVENTIONS 
 Two studies compared oral steroid with placebo (Blockey 1954; Buchbinder 2004). In the first trial the oral steroid consisted of 50 mg cortisone acetate (made upon in a suspension with a concentration of 12.5 mg per ml) four times daily for 3 days (daily dose 200 mg, equivalent prednisolone dose= 8 mg), 25 mg four times daily for 11 days (100 mg daily dose, equivalent prednisolone dose= 4 mg) and then the daily dose was tapered in 12.5 mg decrements (or equivalent to 0.5 mg prednisolone every 2 days (total dose given over 4 weeks = 2.5 g, equivalent prednisolone total dose = 100 mg) and the placebo was a similar inert suspension (Blockey 1954). All patients were instructed to exercise their shoulders vigorously after a week. At the end of 4 weeks patients who had not progressed satisfactorily had their shoulders manipulated under general anaesthesia and were then given a second course of either oral steroid or placebo. The second trial administered 30 mg prednisolone (6 x 5 mg tablets) as a single morning dose for 3 weeks or identical placebo tablets (Buchbinder 2004). All participants also received a simple exercise program comprising pendular exercises and scapular setting (isometric scapular retraction).

One trial compared oral steroid (10 mg prednisolone given in the morning for 4 weeks followed by 5 mg for 2 weeks) with no treatment (Binder 1986). All participants were taught home pendular exercises and were advised to do them for 2‐3 minutes every hour.

One trial compared oral steroid (triamcinolone 4 mg tablets (equivalent prednisolone dose = 5 mg), given 3 times daily for a week, 2 times daily for a week and once daily for a week) with one intra‐articular steroid injection (40 mg triamcinolone acetonide)(Widiastuti‐Samekto). All patients also received physiotherapy commencing on the fourth day and consisting of 12 sessions of 20 minutes of active exercise and passive joint mobilisation treatment.

The remaining study compared oral steroids with placebo prior to and following manipulation under anaesthesia combined with injection of hydrocortisone acetate 25 mg and 0.5% bupivacaine 10 ml into the glenohumeral joint. The oral steroid consisted of 5 mg prednisone 3 times daily for 2 weeks (equivalent to same dose of prednisolone) prior to manipulation and in diminishing doses for 2 weeks following manipulation (exact schedule not stated)(Kessel 1981). All patients also received supervised physiotherapy consisting of graded exercises.

STUDY PARTICIPANTS 
 Nearly all trials recruited similar study populations (See Table of Characteristics of Included Studies). The duration of symptoms was 1‐12 months in Binder et al (Binder 1986) with a mean duration of 5.5 months. In Blockey et al (Blockey 1954) the mean duration of symptoms was 5.1 months in the oral prednisolone group and 6.1 months in the no treatment group with 15/16 participants in each treatment group having a duration of symptoms of 1 year or less. Participants in the study by Buchbinder et al (Buchbinder 2004) had a mean duration of symptoms of 25.5 and 21.1 weeks in the oral prednisolone and placebo groups respectively. Kessel et al (Kessel 1981) did not specify the duration of symptoms. Widiastuti‐Samekto and Sianturi (Widiastuti‐Samekto) reported duration of symptoms of 1‐6 months.

Mean age in the three trials that reported this information was similar: 54.8 years (range 45‐76 years) (Binder 1986); 58 years (range 45‐70 years) and 52 years (range 45‐60 years) in the oral steroid and control groups respectively (Blockey 1954); 53.5 years (SD 5.1) and 55.0 years (SD 9.0) in the oral steroid and placebo groups respectively (Buchbinder 2004). Widiastuti‐Samekto and Sianturi (Widiastuti‐Samekto) reported an age range of 40‐69 years and the age of participants was not stated in Kessel et al (Kessel 1981). There was a female predominance in 3 trials: 24 female and 16 males (Binder 1986); 35 females and 14 males (Buchbinder 2004); 16 females and 13 males (Widiastuti‐Samekto). There were 18 males and 14 females in one trial (Blockey 1954) and the gender of participants was not reported in one trial (Kessel 1981).

TIMING OF FOLLOW UP 
 Follow up assessments were performed at varying time points across the trials, from during treatment to 6 months after the end of treatment. Binder et al performed follow up assessments at 2 and 4 weeks following the commencement of treatment, at 6 weeks (coinciding with the end of treatment) as well as monthly for the next 6 months following treatment (Binder 1986). Blockey et al performed assessments at 1 week following the commencement of treatment (oral prednisolone or placebo), at 4 weeks (coinciding with the end of the first 4 weeks of oral prednisolone or placebo), and at 5, 8 and 18 weeks following commencement of treatment (Blockey 1954). Buchbinder et al performed assessments at 3, 6 and 12 weeks following the commencement of treatment (Buchbinder 2004). The timing of assessments was unclear in Kessel et al but data were presented for 6, 12 and 18 weeks (Kessel 1981). Widiastuti‐Samekto and Sianturi (Widiastuti‐Samekto) assessed outcomes once per week for three weeks.

OUTCOME ASSESSMENT 
 Only one trial specified a primary endpoint: overall pain at three weeks (Buchbinder 2004). Four trials included pain scales but these varied between studies: pain at night, pain on movement and pain at rest on 10 cm visual analogue scale (Binder 1986); spontaneous pain and pain on movement measured on a 4 point categorical scale where 0=no pain, 1=slight pain, 2=moderate pain and 3=severe pain (Blockey 1954); overall pain, night pain, activity related pain on vertical Likert scales from 0‐10 where 0=no pain and 10=maximal imaginable pain (Buchbinder 2004); pain on a visual analogue scale, without specifying any further details of the scale (Widiastuti‐Samekto).

All trials measured shoulder range of movement but the movements and how they were measured varied between trials. Binder et al measured passive movements (total flexion, glenohumeral flexion, total abduction, glenohumeral abduction, total rotation and external rotation using a hydrogoniometer (Binder 1986). Binder et al also calculated a "principal component" termed "C" which consisted of (0.506 x TF) + (0.215 x GF) + (0.583 x TA) + (0.253 x GA) + (0.124 x ER) + (0.528 x TR) ‐ 163.39. Blockey et al measured total abduction, glenohumeral abduction and rotation at the shoulder joint, but did not specify the method of measurement (Blockey 1954). Buchbinder et al measured active range of movement using an inclinometer to measure total shoulder flexion, total shoulder abduction and external rotation in neutral. Hand behind the back was measured by assessment of the anatomical landmark reached by the extended thumb (Buchbinder 2004). Kessel et al measured external rotation and elevation in flexion and expressed these as a percentage of normal as judged by the opposite shoulder (Kessel 1981). Widiastuti‐Samekto and Sianturi measured range of motion to determine which participants met the definition of cured (defined as able to achieve 90% of normal range of motion for abduction and external rotation)(Widiastuti‐Samekto).

Buchbinder et al also measured function and quality of life using a variety of validated questionnaires including the Shoulder Pain and Disability Index (SPADI), Croft shoulder disability questionnaire, Disabilities of Arm, Shoulder and Hand (DASH) functional assessment, the Health Assessment Questionnaire (HAQ) and the Short Form 36 health survey (SF‐36) (Buchbinder 2004). Any adverse effects were also reported.

Risk of bias in included studies

All included studies were small with sample sizes ranging from 28 to 49 participants. While all trials were described as randomised, only three trials described their method of randomisation: computer‐generated permuted block randomisation (Buchbinder 2004); a pre‐arranged random order, devised separately for those with symptoms for more than or less than six months (method for generating the random order not provided)(Blockey 1954); and generation of a random sequence using random number tables (Widiastuti‐Samekto).

Two trials reported adequate concealment of treatment allocation (Blockey 1954; Buchbinder 2004) and three were unclear (Binder 1986; Kessel 1981; Widiastuti‐Samekto). Participants were blinded in three trials (Blockey 1954; Buchbinder 2004; Kessel 1981) and outcome assessors were blinded to treatment allocation in all trials. Three trials reported that the analysis was performed on the basis of intention to treat (Binder 1986; Blockey 1954; Buchbinder 2004) whereas the method of analysis was unclear in two studies (Kessel 1981; Widiastuti‐Samekto). Only one trial reported a sample size calculation (Buchbinder 2004). A summary of the methodological quality assessment of each trial is provided in the Characteristics of Included Studies Table.

Effects of interventions

Data from the five trials could not be pooled due to different comparator arms in 3 trials (Binder 1986; Kessel 1981; Widiastuti‐Samekto); and inability to extract any raw data from one of the two placebo‐controlled trials (as only mean data without any measures of variance were reported)(Blockey 1954).

EFFICACY 
 ORAL STEROIDS VERSUS PLACEBO 
 One trial of 30 participants reported no significant differences between groups in terms of pain or range of movement (Blockey 1954) although there was a suggestion that improvement in both pain and range of movement occurred earlier in the steroid treated group (see Additional Tables, Table 1; Table 2). A second trial of 49 participants reported significantly greater improvements for participant reported success (RR = 2, 95% CI 1.3 to 3.1), pain (e.g. overall pain improved by 2.7 (95% CI 1.4 to 4.0) on a 0 to 10 point scale), range of movement (e.g. total shoulder abduction increased by 23.3 degrees (95% CI 11.3 to 35.5) and shoulder‐specific disability (e.g. SPADI improved by 18.1 (95% 7.6 to 28.6) on a 0 to 100 point scale at 3 weeks in the prednisolone group compared with the placebo group (see Comparisons and Data Analyses 01, and Additional Tables, Table 3; Table 4)(Buchbinder 2004). There were no differences between groups at six or 12 weeks. The differences between treatment groups was not constant over time with improvements in the prednisolone group being maximal at 3 weeks after which time the participants remained stable or deteriorated somewhat and in contrast there was a small improvement in the placebo group at 3 weeks after which time the participants continued to improve at each time point with maximal improvement noted at 12 weeks.

1. Blockey 1954 Mean Pain over time (0:none 1:slight 2:moderate, 3:severe).

Intervention	Mean Pain (baseline)	Mean Pain (1 weeks)	Mean pain (4 weeks)	Mean Pain (18 weeks)
Control group	1.4	1.3	0.8	0.5
Steroid Group	1.4	0.9	0.5	0.6

2. Blockey 1954 Range of movement over time.

Intervention	Baseline (degrees)	Improvement (week 1)	Improvement (week 4)	Improvement (wk 18)
Control group
Total abduction	75	+14	+17	+48
Shoulder joint abduction	36	+7	+3	+33
Shoulder joint rotation	40	+7	+10	+53
Steroid group
Total abduction	82	+21	+22	+28
Shoulder joint abduction	42	+12	+9	+13
Shoulder joint rotation	54	+17	+19	+28

3. Clinical relevance table‐ Patient reported success at 3 weeks (Buchbinder 2004).

Outcome	#patients (#trials)	Control event rate	Absolute RD	Rel %change	NNT(B)	Stat significance	Quality of evidence
Pt reported success	49 (1)	48% 48 out of 100	48% 48 more out of 100	100% (I)	2	significant	silver
Legend Pt=patient			RD=risk difference		NNT=number needed to treat to benefit

4. Clinical relevance table‐ Pain, SPADI, shoulder flexion and abduction at 3 weeks.

Outcome	#patients (#trials)	Control baseline m*	Absolute change	Rel %change	NNT	Stat significance	Quality of evidence
Overall pain (3 weeks) 0‐10 VAS	49(1)	6.8	27% 2.7 points on 10 point scale	40% (I)	2	significant	silver
SPADI score (3 weeks) (0‐100)	49(1)	60.5	18% 18 points on 100‐point scale	30% (I)	3	significant	silver
Total shoulder flexion (3 weeks) (0‐180 degrees)	49(1)	93.2	11% 20.3 degrees on 180 degree scale	22% (I)		significant	silver
Total shoulder abduction (3 weeks) (0‐180 degrees)	49(1)	70.2	13% 23.3 degrees on 180 degree scale	33% (I)		significant	silver
		*m=mean (placebo)		I=improvement

ORAL STEROIDS VERSUS NO TREATMENT 
 One trial of 40 participants reported that the oral steroid group had a more rapid initial recovery in terms of pain but by five months the difference between the groups was negligible (Binder 1986). There was no significant difference between the oral steroid and no treatment groups in terms of number of participants left with residual overall pain, residual pain on movement, residual night pain or range of movement (flexion, abduction and external rotation) at 8 months follow‐up (see Comparisons and Data Analyses 02).

ORAL STEROIDS, MANIPULATION UNDER ANAESTHESIA AND STEROID INJECTION VERUS PLACEBO, MANIPULATION AND STEROID INJECTION 
 One trial of 32 participants reported no difference between groups in the number of participants who responded dramatically to manipulation under anaesthesia (RR 1.87, 95% CI 0.78 to 4.46)(Kessel 1981) (see Comparisons and Data Analyses 03). External rotation was significantly better in those receiving oral steroids prior to manipulation compared to those receiving placebo prior to manipulation at 6 weeks follow up (RR 4.67, 95% CI 1.17, 18.58) but there were no differences between treatment groups at 12 or 18 weeks. Shoulder flexion was not significantly different at any time point (6, 12 or 18 weeks).

ORAL STEROIDS VERSUS INTRA‐ARTICULAR CORTICOSTEROID INJECTION 
 One trial of 28 participants found a better cure rate (defined as able to achieve 90% of normal range of glenohumeral motion for abduction and external rotation) in the steroid injection group compared to oral steroids after one week (RR = 0.22 (95% CI 0.06 to 0.84)) but this was not statistically significant at weeks 2 or 3 (RR=0.65 (95% CI 0.42 to 1.01) and RR=0.80 (95% CI 0.62 to 1.03) respectively) (Comparisons and Data Analyses 04)(Widiastuti‐Samekto). A significant difference in pain between the two groups was also reported after one week although which group had better scores was not reported, and there were no differences between groups at weeks 2 or 3.

ADVERSE EFFECTS 
 Three studies report adverse events. One trial reported no statistically significant difference in the number of reported adverse events in the oral steroid group compared to the placebo group and no serious events occurred in either group (Buchbinder 2004). One participant in the placebo group developed a stress fracture in the foot. Widiastuti‐Samekto and Sianturi reported that three participants in the oral steroid group had epigastric pain and three participants in the steroid injection group had pain at the injection site (Widiastuti‐Samekto). Blockey et al reported that one participant in the oral steroid group died suddenly from a coronary artery occlusion during the third week of treatment. His past history and risk of heart disease are not described but the authors stated that this was probably not attributable to steroid treatment (Blockey 1954). A second participant, also in the active group, developed follicular dermatitis during the fourth week of treatment and subsequently withdrew from the trial.

Discussion

It is not possible to draw firm conclusions about the efficacy of oral steroids for adhesive capsulitis from this review although the available data is suggestive of worthwhile short‐term benefits. Two placebo‐controlled trials including 79 participants were included in this systematic review (Blockey 1954; Buchbinder 2004). Data from one trial demonstrated that a three week course of 30 mg prednisolone daily is of significant short‐term benefit for adhesive capsulitis but the benefits are not maintained at 6 weeks (Buchbinder 2004) while the second trial of 30 participants reported no significant differences between groups in terms of pain or range of movement although there was a suggestion that improvement occurred earlier in the steroid treated group using a tapering dose of cortisone acetate over 4 weeks (total dose given over 4 weeks = 2.5 g)(Blockey 1954). A more rapid improvement in pain was also demonstrated in a third trial of 40 participants that compared 10 mg prednisolone given in the morning for 4 weeks followed by 5 mg for 2 weeks to no treatment (Binder 1986). It appears from these trials that the effect of oral steroids may not be constant over time, but that the treatment effect may diminish over time. However, two trials also noted a rebound effect upon cessation of oral steroids (Binder 1986; Buchbinder 2004) suggesting that significant improvement may have continued and increased with a more prolonged course of therapy and/or more gradual withdrawal of treatment. No conclusions can be drawn about the comparative efficacy of oral versus intra‐articular steroid or the additional benefit of oral steroids over manipulation under anaesthesia and intra‐articular steroid injection.

The adverse effects of steroid therapy in the two of the reviewed trials that reported adverse effects in a systematic manner were minor and short‐lived although a valid estimate of clinically significant uncommon or rare adverse effects cannot be made from these trials. A case of sudden death attributed to coronary artery occlusion and one case of follicular dermatitis was reported in a third trial. The maximal daily dose of oral steroids in this trial was the equivalent of only 8 mg prednisolone. The adverse effects of steroid therapy, particularly in the setting of long term use in rheumatic diseases have been well described (Da Silva 2006). A recent review of adverse effects of corticosteroids on the cardiovascular system concluded that the major adverse effects are dyslipidaemia and hypertension (Sholter 2000). These may predispose treated patients to coronary artery disease if high doses and prolonged courses are used. Another recent review supports this view and found that overall, evidence is lacking that low dose glucocorticoid therapy (less than or equal to 10 mg prednisolone daily) significantly increases the incidence of cardiovascular disease in rheumatoid arthritis (in contrast with higher doses)(Da Silva 2006). Blockey et al suggested that the death in their trial was probably not attributable to oral steroid treatment since there is a known association between adhesive capsulitis and coronary artery disease (Blockey 1954). This association has not been verified although a small case‐control study in 1997 did report that both diabetes and heart disease appeared to be more prevalent in patients diagnosed with adhesive capsulitis as compared with a control group of patients attending the same facility without the condition and compared with the general population (Boyle‐Walker 1997).

We have previously examined the selection criteria used in trials of adhesive capsulitis (Green 1998b). While most study populations were broadly able to be categorised as adhesive capsulitis (which included frozen shoulder and periarthritis) based upon the diagnostic labels and/or definitions of these labels when described, some trials did not specify a diagnosis and some trials gave no selection criteria or study population definition. In this review of five RCTs, one trial did not specify any specific inclusion criteria that would indicate adhesive capsulitis (only specified 'periarthritis in one or both shoulders') (Blockey 1954), while the other four trials all required pain and varying degrees of restriction of passive glenohumeral movements in at least two planes of movement (Binder 1986; Buchbinder 2004; Kessel 1981; Widiastuti‐Samekto) although none of the criteria were exactly the same. The development of standard criteria to define adhesive capsulitis and other shoulder disorders would ensure the homogeneity of study populations across studies.

It has been postulated that oral steroids may be more effective in the earlier phase of adhesive capsulitis yet most trials in this review recruited participants with a median duration of symptoms of 6 months. While limiting participation in trials of corticosteroids to those with recent onset of symptoms may appear to have merit, early recruitment has proven universally difficult for trialists in this field (Buchbinder 2004) and may not be achievable.

All trials included in this review were of small size, ranging from 28 to 49 participants, although one trial (49 participants) determined a priori that this sample size would have adequate power to determine a clinically important difference if one was present (Buchbinder 2004). This was the only trial to be considered of high methodological quality. The other trials were of variable quality and some were poorly reported. In addition none of the trials used the same oral steroids regimen. This may reflect changes in standard practices over time. In future trials, the CONSORT statement should be used as a guide for both designing and reporting trials (www.consort‐statement.org). Trial reporting should include the method of randomisation and treatment allocation concealment, follow up of all participants who entered the trial, and an intention to treat analysis. Sample sizes should be reported and have adequate power to answer the research question, and ideally trials should include both short‐term and long‐term follow‐up. To enable comparison and pooling of the results of RCTs, we suggest that future trials report means with standard deviations for continuous measures or number of events and total numbers analysed for dichotomous measures.

Further well designed trials are needed to confirm the suggestion from this review that oral steroids provide worthwhile short‐term benefits. It may be that more sustained benefits can be achieved by a more prolonged course of treatment and/or a more gradual tapering of the dose (without concomitantly increasing the risk of side effects) and/or combination with other efficacious treatments. While the adverse effects of steroid therapy in the trials in the review were minor and short‐lived, the potential risks of oral steroids are well described. Therefore, the potential benefits and risks should be weighed up in individual patients, particularly those with known risk factors for the development of adverse effects.

Authors' conclusions

Implications for practice.

It is not possible to draw firm conclusions about the efficacy of oral steroids for adhesive capsulitis from this review although the available data is suggestive of worthwhile short‐term benefits. There is Silver level evidence based upon three small randomised controlled trials that a short course of oral steroids for adhesive capsulitis may be of significant short‐term benefit compared to placebo or no treatment but the effect is not maintained at six weeks. There is a lack of reliable evidence to support or refute the efficacy of oral steroids compared to other treatments (specifically, intra‐articular corticosteroid injection alone or in combination with manipulation under anaesthesia). While the adverse effects of steroid therapy in the trials in the review were minor and short‐lived, the potential risks of oral steroids are well described. Therefore, the potential benefits and risks should be weighed in individual patients.

Implications for research.

Further well designed trials are needed to confirm the short‐term benefits of oral steroids and determine whether more sustained benefits can be achieved by a more prolonged course of treatment and/or a more gradual tapering of the dose (without concomitantly increasing the risk of side effects) and/or combination with other efficacious treatments. The CONSORT statement should be used as a guide for both designing and reporting trials (www.consort‐statement.org). Trial reporting should include the method of randomisation and treatment allocation concealment, follow up of all participants who entered the trial, and an intention to treat analysis. Sample sizes should be reported and have adequate power to answer the research question, and ideally trials should include both short‐term and long‐term follow‐up. To enable comparison and pooling of the results of RCTs, we suggest that future trials report means with standard deviations for continuous measures or number of events and total numbers analysed for dichotomous measures. Development of a standard set of outcome measures including a definition of what constitutes a clinically important improvement for adhesive capsulitis would significantly enhance these research endeavours.

What's new

Date	Event	Description
7 November 2008	Amended	Converted to new review format. CMSG ID: C059‐R

History

Review first published: Issue 4, 2006

Date	Event	Description
6 August 2006	Amended	Oral steroids for adhesive capsulitis is one of several interventions that was originally included in the following Cochrane review which included all interventions for shoulder pain: Green S, Buchbinder R, Glazier R, Forbes A. Interventions for shoulder pain. The Cochrane Database of Systematic Reviews 1999, Issue 2. Art. No.: CD001156. DOI: 10.1002/14651858.CD001156. Since the original review, many new clinical trials, studying a diverse range of intervetions, have been performed. In order to update the review, it has been subdivided into a series of reviews investigating the evidence for efficacy of single interventions.

Appendices

Appendix 1. MEDLINE search strategy

1. randomized controlled trial.pt. 
 2. controlled clinical trial.pt. 
 3. randomized controlled trials.sh. 
 4. random allocation.sh. 
 5. double blind method.sh. 
 6. single blind method.sh. 
 7. or/1‐6 
 8. (animals not human).sh. 
 9. 7 not 8 
 10. clinical trial.pt. 
 11. exp clinical trials/ 
 12. (clin$ adj25 trial$).ti,ab. 
 13. ((singl$ or doubl$ or trebl$ or tripl$) adj25 (blind$ or mask$)).ti,ab. 
 14. placebos.sh. 
 15. placebo$.ti,ab. 
 16. random$.ti,ab. 
 17. research design.sh. 
 18. or/10‐17 
 19. 18 not 8 
 20. 19 not 9 
 21. 9 or 20 
 22. exp shoulder pain/ 
 23. (shoulder adj5 pain$).tw. 
 24. exp rotator cuff/ 
 25. (rotator adj5 cuff).tw. 
 26. exp Bursitis/ 
 27. adhesive capsulitis.tw. 
 28. exp Periarthritis/ 
 29. (periarthritis adj5 shoulder).tw. 
 30. (frozen adj5 shoulder).tw. 
 31. (capsulitis adj5 shoulder).tw. 
 32. or/22‐31 
 33. exp CORTISONE/ 
 34. exp PREDNISONE/ 
 35. exp PREDNISOLONE/ 
 36. exp Glucocorticoids/ 
 37. (cortisone or prednisone or prednisolone or oral glucocorticosteroid$ or glucocortisone$).tw. 
 38. or/33‐37 
 39. 21 and 32 and 38

Appendix 2. Search strategies ‐ other databases

CENTRAL	EMBASE	CINAHL
1. (shoulder near pain) 2. (rotator near cuff) 3. (adhesive capsulitis) 4. (periarthritis near shoulder) 5. (frozen shoulder) 6. or/1‐5 7. glucocorti* or cortisone* or predniso* or (oral steroid*) 8. 6 and 7	1. 'randomized controlled trial'/exp OR 'clinical trial'/exp OR 'double blind procedure'/exp OR 'single blind procedure'/exp 2. random*:ti,ab OR cross?over*:ti,ab OR factorial*:ti,ab OR placebo*:ti,ab 3. 'shoulder pain'/exp OR 'humeroscapular periarthritis'/exp OR 'frozen shoulder'/exp OR 'bursitis'/exp 4. 'adhesive capsulitis' 5. 'frozen shoulder' OR 'shoulder pain' AND [embase]/ 6. 'corticosteroid'/exp OR 'prednisolone'/exp OR 'prednisone'/exp 7. predniso* OR glucocorti* OR cortisone* 8. (#1 OR #2) AND (#3 OR #4 OR #5) AND (#6 OR #7)	1. clinical trial.pt. 2. exp clinical trials/ 3. (clin$ adj25 trial$).ti,ab. 4. ((singl$ or doubl$ or trebl$ or tripl$) adj25 (blind$ or mask$)).ti,ab. 5. placebos.sh. 6. placebo$.ti,ab. 7. random$.ti,ab. 8. research design.sh. 9. or/1‐8 10. shoulder pain.mp. [mp=title, subject heading word, abstract, instrumentation] 11. exp Shoulder Pain/ 12. (shoulder adj5 pain).mp. 13. exp Adhesive Capsulitis/ 14. (adhesive adj5 capsulitis).mp. 15. exp bursitis/ or exp adhesive capsulitis/ 16. (bursitis adj5 shoulder).mp. 17. periarthritis.mp. 18. (periarthritis adj5 shoulder).mp. [mp=title, subject heading word, abstract, instrumentation] 19. (frozen adj5 shoulder$).mp. 20. (capsulitis adj5 shoulder$).mp. [mp=title, subject heading word, abstract, instrumentation] 21. exp Rotator Cuff Injuries/ 22. (rotator adj5 cuff).mp. 23. or/10‐22 24. Cortisone/ 25. Prednisone/ 26. exp prednisolone/ or exp methylprednisolone/ 27. exp Glucocorticoids/ 28. (glucocorti$ or cortisone$ or predniso$).mp. 29. or/22‐28 30. 23 and 29 31. 9 and 30

Data and analyses

Comparison 1. ORAL STEROID VS PLACEBO.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Patient reported success	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
1.1 3 weeks	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
1.2 6 weeks	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
1.3 12 weeks	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
2 Mean change in overall pain	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
2.1 3 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
2.2 6 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
2.3 12 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
3 Mean change in night pain	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
3.1 3 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
3.2 6 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
3.3 12 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
4 Mean change in movement pain	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
4.1 3 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
4.2 6 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
4.3 12 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
5 Mean change in SPADI	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
5.1 3 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
5.2 6 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
5.3 12 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
6 Mean change in Croft	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
6.1 3 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
6.2 6 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
6.3 12 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
7 Mean change in DASH	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
7.1 3 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
7.2 6 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
7.3 12 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
8 Mean change in HAQ	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
8.1 3 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
8.2 6 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
8.3 12 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
9 Mean change in total shoulder abduction	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
9.1 3 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
9.2 6 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
9.3 12 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
10 Mean change in total shoulder flexion	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
10.1 3 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
10.2 6 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
10.3 12 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
11 Mean change in external rotation in neutral	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
11.1 3 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
11.2 6 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
11.3 12 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
12 Mean change in hand behind back	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
12.1 3 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
12.2 6 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
12.3 12 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
13 Mean change in physical function domain of SF‐36	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
13.1 3 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
13.2 6 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
13.3 12 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
14 Mean change in social function domain of SF‐36	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
14.1 3 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
14.2 6 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
14.3 12 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
15 Mean change in role emotional domain of SF‐36	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
15.1 3 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
15.2 6 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
15.3 12 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
16 Mean change in bodily pain domain of SF‐36	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
16.1 3 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
16.2 6 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
16.3 12 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
17 Mean change in mental health domain of SF‐36	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
17.1 3 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
17.2 6 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
17.3 12 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
18 Mean change in vitality domain of SF‐36	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
18.1 3 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
18.2 6 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
18.3 12 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
19 Mean change in general health domain of SF‐36	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
19.1 3 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
19.2 6 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
19.3 12 weeks	1		Mean Difference (IV, Fixed, 95% CI)	0.0 [0.0, 0.0]
20 Number of patients with one or more reported adverse effects	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
20.1 during treatment period (0‐3 weeks)	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
20.2 after cessation treatment (3‐12 weeks)	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]

1.1. Analysis.

Comparison 1 ORAL STEROID VS PLACEBO, Outcome 1 Patient reported success.

1.2. Analysis.

Comparison 1 ORAL STEROID VS PLACEBO, Outcome 2 Mean change in overall pain.

1.3. Analysis.

Comparison 1 ORAL STEROID VS PLACEBO, Outcome 3 Mean change in night pain.

1.4. Analysis.

Comparison 1 ORAL STEROID VS PLACEBO, Outcome 4 Mean change in movement pain.

1.5. Analysis.

Comparison 1 ORAL STEROID VS PLACEBO, Outcome 5 Mean change in SPADI.

1.6. Analysis.

Comparison 1 ORAL STEROID VS PLACEBO, Outcome 6 Mean change in Croft.

1.7. Analysis.

Comparison 1 ORAL STEROID VS PLACEBO, Outcome 7 Mean change in DASH.

1.8. Analysis.

Comparison 1 ORAL STEROID VS PLACEBO, Outcome 8 Mean change in HAQ.

1.9. Analysis.

Comparison 1 ORAL STEROID VS PLACEBO, Outcome 9 Mean change in total shoulder abduction.

1.10. Analysis.

Comparison 1 ORAL STEROID VS PLACEBO, Outcome 10 Mean change in total shoulder flexion.

1.11. Analysis.

Comparison 1 ORAL STEROID VS PLACEBO, Outcome 11 Mean change in external rotation in neutral.

1.12. Analysis.

Comparison 1 ORAL STEROID VS PLACEBO, Outcome 12 Mean change in hand behind back.

1.13. Analysis.

Comparison 1 ORAL STEROID VS PLACEBO, Outcome 13 Mean change in physical function domain of SF‐36.

1.14. Analysis.

Comparison 1 ORAL STEROID VS PLACEBO, Outcome 14 Mean change in social function domain of SF‐36.

1.15. Analysis.

Comparison 1 ORAL STEROID VS PLACEBO, Outcome 15 Mean change in role emotional domain of SF‐36.

1.16. Analysis.

Comparison 1 ORAL STEROID VS PLACEBO, Outcome 16 Mean change in bodily pain domain of SF‐36.

1.17. Analysis.

Comparison 1 ORAL STEROID VS PLACEBO, Outcome 17 Mean change in mental health domain of SF‐36.

1.18. Analysis.

Comparison 1 ORAL STEROID VS PLACEBO, Outcome 18 Mean change in vitality domain of SF‐36.

1.19. Analysis.

Comparison 1 ORAL STEROID VS PLACEBO, Outcome 19 Mean change in general health domain of SF‐36.

1.20. Analysis.

Comparison 1 ORAL STEROID VS PLACEBO, Outcome 20 Number of patients with one or more reported adverse effects.

Comparison 2. ORAL STEROID TREATMENT VS NO TREATMENT.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Severe residual pain at 8 months	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
2 Mild pain at night at 8 months	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
3 Mild pain on movement at 8 months	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
4 Range of abduction at 8 months	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
5 Range of flexion at 8 months	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected
6 Range of external rotation at 8 months	1		Mean Difference (IV, Fixed, 95% CI)	Totals not selected

2.1. Analysis.

Comparison 2 ORAL STEROID TREATMENT VS NO TREATMENT, Outcome 1 Severe residual pain at 8 months.

2.2. Analysis.

Comparison 2 ORAL STEROID TREATMENT VS NO TREATMENT, Outcome 2 Mild pain at night at 8 months.

2.3. Analysis.

Comparison 2 ORAL STEROID TREATMENT VS NO TREATMENT, Outcome 3 Mild pain on movement at 8 months.

2.4. Analysis.

Comparison 2 ORAL STEROID TREATMENT VS NO TREATMENT, Outcome 4 Range of abduction at 8 months.

2.5. Analysis.

Comparison 2 ORAL STEROID TREATMENT VS NO TREATMENT, Outcome 5 Range of flexion at 8 months.

2.6. Analysis.

Comparison 2 ORAL STEROID TREATMENT VS NO TREATMENT, Outcome 6 Range of external rotation at 8 months.

Comparison 3. ORAL STEROID, MANIPULATION AND STEROID INJECTION VS PLACEBO, MANIPULATION AND STEROID INJECTION.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Dramatic response to manipulation	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
2 External rotation: Better than 3/4 normal movement compared with unaffected shoulder	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
2.1 6 weeks	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
2.2 12 weeks	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
2.3 18 weeks	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
3 Flexion: Better than 3/4 normal movement compared with unaffected shoulder	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
3.1 6 weeks	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
3.2 12 weeks	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
3.3 18 weeks	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]

3.1. Analysis.

Comparison 3 ORAL STEROID, MANIPULATION AND STEROID INJECTION VS PLACEBO, MANIPULATION AND STEROID INJECTION, Outcome 1 Dramatic response to manipulation.

3.2. Analysis.

Comparison 3 ORAL STEROID, MANIPULATION AND STEROID INJECTION VS PLACEBO, MANIPULATION AND STEROID INJECTION, Outcome 2 External rotation: Better than 3/4 normal movement compared with unaffected shoulder.

3.3. Analysis.

Comparison 3 ORAL STEROID, MANIPULATION AND STEROID INJECTION VS PLACEBO, MANIPULATION AND STEROID INJECTION, Outcome 3 Flexion: Better than 3/4 normal movement compared with unaffected shoulder.

Comparison 4. ORAL VS INTRA‐ARTICULAR STEROID.

Outcome or subgroup title	No. of studies	No. of participants	Statistical method	Effect size
1 Number cured (90% improved in glenohumeral abduction and external rotation)	1		Risk Ratio (M‐H, Fixed, 95% CI)	Totals not selected
1.1 1 week	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
1.2 2 weeks	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]
1.3 3 weeks	1		Risk Ratio (M‐H, Fixed, 95% CI)	0.0 [0.0, 0.0]

4.1. Analysis.

Comparison 4 ORAL VS INTRA‐ARTICULAR STEROID, Outcome 1 Number cured (90% improved in glenohumeral abduction and external rotation).

Characteristics of studies

Characteristics of included studies [ordered by study ID]

Binder 1986.

Methods	Randomised controlled trial. Randomisation method not described Blinding: outcome assessment was blinded but participants were not. Loss to follow‐up: zero Sample size calculation: not reported Appropriate statistical analysis: yes, intention to treat analysis.
Participants	40 participants, 24 females and 16 males Inclusion criteria: painful stiff shoulders of at least 1 month's duration; pain with sleep disturbance; restricted active and passive shoulder movement with a reduction in external rotation of at least 50% Exclusion criteria: generalised arthritis; sensory symptoms or signs in the arm; radiation of pain to the neck; peptic ulceration; serious infection; contraindications to systemic steroid therapy.
Interventions	Group 1(20 participants): 10mg enteric coated prednisolone as a morning dose for 4 weeks, then 5mg per day for 2 weeks. Group 2 (20 participants): No treatment All participants in both groups were taught home pendular exercises and advised to do them for 2‐3 minutes every hour. Nonsalicylate analgesics and diazepam 5 mg at night were available if requested but NSAIDs were stopped.
Outcomes	Assessed at baseline, 2, 4, 6 weeks and monthly for a further 6 months. 1) Pain at night, pain on movement and pain at rest during the day on 10cm VAS 2) Passive movements were measured using a hydrogoniometer a. Total flexion (TF) b. Glenohumeral flexion (GF) c. Total abduction (TA) d. Glenohumeral abduction (GA) e. Total rotation (TR) f. External rotation (ER) Calculated a "principal component" (C) from these measures = (0.506 x TF) + (0.215 x GF) + (0.583 x TA) + (0.253 x GA) + (0.124 x ER) + (0.528 x TR) ‐ 163.39.
Notes	Pain was measured on visual analogue scales (VAS) but reported as number of participants with residual pain at 8 months. Only means and range of movement at 8 months was reported (authors report that baseline measurements were comparable between groups). Number of participants with residual pain at 8 months and range of movement at 8 months was able to be used for meta‐analysis (by computing standard deviations from mean and range scores by assuming standard deviation equals one quarter of the range). All other data was presented graphically, without any measures of variance and therefore could not be used for meta‐analysis. Authors reported that the pattern of improvement in pain at night over 8 weeks showed a significant difference in favour of the oral steroid group with a more rapid initial recovery, although by 5 months the difference between groups was negligible. Improvements in pain at rest and with movement, range of motion and a cumulative recovery curve were not significantly different between groups over 8 months. No external funding acknowledgements listed.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	B ‐ Unclear

Blockey 1954.

Methods	Randomised controlled trial. Randomisation: allocated to treatment on the basis of a pre‐arranged random order stratified for duration of symptoms more than or less than 6 months. The treatment register was held by the hospital pharmacist, and the 2 surgeons admitting participants to the trial and assessing their progress did not know which participants were receiving cortisone until after the whole trial had been completed. Blinding: both participants and outcome assessors were blinded. Loss to follow‐up: 2 (12.5%) patients from cortisone group: a 69 year old man who died suddenly from coronary occlusion during the third week of treatment and a 53 year old woman who developed follicular dermatitis during the fourth week of treatment and was withdrawn. Sample size calculation: not reported Appropriate statistical analysis: yes, intention to treat analysis.
Participants	32 participants Inclusion criteria: Periarthritis of one or both shoulders; aged between 20 and 70yrs old; no symptoms or signs of disease in other joints; no evidence of arthritis or bone diseases on x‐ray; blood sedimentation rate (Wintrobe) under 20mm in first hour; chest x‐rays within normal limits. Exclusion criteria: inflammatory conditions; coronary disease
Interventions	Group 1(16 participants): cortisone acetate suspension in syrup (concentration = 12.5mg/ml) taken orally in four divided doses during the day. 200mg for 1st 3 days, and 100mg thereafter till 14th day. The daily dose was then tailed off in decrements in 12.5mg every 2 days. The complete dosage for four weeks = 2.5g Group 2 (16 participants): placebo inert suspension in similar dosage. All participants in both groups were instructed to exercise their shoulders vigourously. At end of 4 weeks, the participants who had not progressed satisfactorily had their shoulders manipulated under general anaesthesia. This manipulation was followed by a 2nd four‐weeks course of cortisone or placebo. All cases followed for 18 weeks.
Outcomes	Assessed at baseline, 1, 4, 5, 8 and 18 weeks 1) Spontaneous pain 2) Pain on movement (Both measured by 0=none, 1=slight, 2=moderate, 3=severe) 3) Total abduction (degrees) 4) Glenohumeral abduction (degrees) 5) Rotation at shoulder joint (degrees).
Notes	Only mean data without any measures of variance were reported and therefore data could not be used for meta‐analysis. 4‐point categorical scale analysed as interval scale. The authors reported that there was no statistically significant difference between the two groups however there was a suggestion that improvement in pain and range of movement occurred earlier in the steroid treated group (see Additional Tables 1 and 2). Steroid therapy also reduced the number of patients requiring manipulation at the end of 4 weeks (11/16 (68.8%) in the control group vs 6/15 (40%) in the steroid group) for failure to improve (not statistically significant). There was one death due to coronary occlusion during the third week of treatment and one participant developed follicular dermatitis in the fourth week of treatment. Both participants were in the oral steroid group. No external funding acknowledgements listed.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Low risk	A ‐ Adequate

Buchbinder 2004.

Methods	Randomised controlled trial Randomisation: computer generated permuted block randomisation of eight. Allocation concealment ensured and study biostatistician kept assignment scheme. Blinding: both participants and outcome assessors were blinded. Loss to follow‐up: No patients prior to 3 week follow up , although 1 was lost to follow up at 6 weeks and 3 were lost to follow up at 12 weeks. Sample size reported: yes. Sample size: 21 participants per group would have 90% power at a significance level of 0.05 to detect a clinically important difference in pain perception of two (on 0 to 10 scale) assuming standard deviation is 2. Appropriate statistical analysis: yes, intention to treat analysis
Participants	49 participants Inclusion criteria: pain and stiffness in predominantly one shoulder for greater than or equal to 3 weeks; restriction of passive motion of greater than 30 degrees in two or more planes of movement measured to onset of pain with a gravity inclinometer; adults > 18 years. Exclusion criteria: systemic inflammatory joint disease (including rheumatoid arthritis, polymyalgia rheumatica); oral steroids in previous 3 months; diabetes mellitus (because of potential for unblinding); pregnancy; contraindications to oral steroids including peptic ulceration; serious infection or uncontrolled hypertension; radiological evidence of osteoarthritis of the shoulder or fracture; calcification about the shoulder joint; reason to suspect a full rotator cuff tear (weakness of arm elevation, a positive "drop arm sign", a high riding humerus visible on x‐ray of the shoulder or demonstration of a complete rotator cuff tear on ultrasound); likely not to comply with follow up (e.g.: living too far away to attend for follow up assessment and/or those indicating they would be unable and/or unwilling to attend for outcome assessment); lack of written informed consent.
Interventions	Group 1 (oral prednisolone group) (24 participants): 6 x 5mg tablets or oral prednisolone daily for 3 weeks as a single morning dose. Group 2 (placebo group) (26 participants): 6 x 5mg placebo tablets (identical in taste and appearance to the prednisolone tablets) daily for 3 weeks as a single morning dose. Participants in both groups received a simple exercise program comprising pendular exercises and scapular setting (isometric scapular retraction). Participants were asked to cease non‐steroidal anti‐inflammatory medication but were allowed paracetamol and codeine preparations. No other interventions were allowed for the duration of the trial. For ethical reasons, if the patients had not improved after 6 weeks, the treating physician could request an unblinding of treatment allocation and further manage the condition at their discretion. This was considered a protocol violation but the participant continued to be followed up and the outcome assessor remained blinded.
Outcomes	Outcome assessed at baseline, 3, 6 and 12 weeks. Primary outcome measure was overall pain perception at 3 weeks. 1) Pain perception: overall pain, night pain, activity related pain on vertical Likert scale from 0‐10 where 0 = no pain and 10 = maximal imaginable pain. 2) Shoulder Pain and Disability Index (SPADI): 13 items divided into 2 subscales (pain ‐ 5 items and disability ‐ 8 items). Score out of 100 where higher scores reflect more pain/disability. 3) Croft shoulder disability questionnaire: score out of 22 where a higher score reflects more disability. 4) Disabilities Arm Shoulder and Hand (DASH) questionnaire: Score expressed as a percentage score (0‐100). 5) Health Assessment Questionnaire (HAQ): arthritis‐specific functional assessment measure. The disability score is expressed between 0 (no disability) and 3 (highest disability). 6) Short Form 36 Health Survey (SF‐36): self‐administered 36 item generic indicator of health status consisting of 8 subscales. Each of the 8 subscales is rescaled from 0‐100 where higher scores represent better health. 7) Participant‐rated improvement compared to baseline on a 5‐point categorical scale (marked improvement, moderate improvement, same, moderate worsening, marked worsening). Success was defined a priori as marked or moderate improvement. 8) Range of motion: total shoulder flexion, total shoulder abduction and external glenohumeral rotation in neutral abduction were measured using an inclinometer and hand behind back was measured by assessment of the anatomical landmark reached by extended thumb. 9) Adverse effects.
Notes	Sigma pharmaceuticals provided the prednisolone tablets at no cost. No other external funding acknowledgements listed.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Low risk	A ‐ Adequate

Kessel 1981.

Methods	Randomised controlled trial. Randomisation: allocated by randomized selection to one of two groups, method not described. Blinding: Both participants and medical staff concerned with treatment blinded (presumed to have performed outcome assessments). Loss to follow up: 5 (15.6%) patients were excluded from the analysis on the basis of insufficient information. Sample size calculation: not reported Appropriate statistical analysis: unclear
Participants	30 participants (5 excluded = 27 patients, 28 shoulders (one patient sequential bilateral involvement during 3 years of trial so included twice) Inclusion criteria: Spontaneous onset of pain localized to the shoulder region with the pain increasing in severity and usually being worse at night; limitation of all total shoulder movements by at least 50%; no clinically or radiologically identifiable lesion of the shoulder.
Interventions	Group 1 (14 patients): prednisone 5 mg three times daily for 2 weeks prior to manipulation and 2 weeks thereafter in diminishing doses. Group 2 (16 patients): identical placebo. Both groups had 2 weeks' rest in sling followed by manipulation under anaesthesia combined with injection of hydrocortisone acetate 25 mg and 0.5% bupivacaine 10 ml into the glenohumeral joint. Following manipulation all patients received supervised physiotherapy consisting of graded exercises.
Outcomes	1) "dramatic response to manipulation" (audible cracking of adhesions and full range of motion obtained under anaesthesia)‐ measured at manipulation under anaesthesia 2) shoulder movements ‐ reported for external rotation and elevation in flexion, expressed as a percentage of normal as judged by the opposite shoulder, (percentage of participants with better than three‐quarter normal movements compared with the normal shoulder); data presented for 6, 12 and 18 weeks post‐manipulation.
Notes	The timing of the outcome assessments was unclear but data was presented for 6, 12 and 18 weeks post manipulation. The method of data analysis was also unclear. The authors reported shoulder movements as percentage at each time point with better than 3/4 normal movements compared with the unaffected shoulder; we converted this to proportion of participants with better than 3/4 normal shoulder movement compared with the unaffected shoulder. No adverse events listed No external funding acknowledgements listed.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	B ‐ Unclear

Widiastuti‐Samekto.

Methods	Randomised controlled trial. Randomisation: participants allocated into two groups using random number tables to generate a random sequence; numbered, sealed envelopes containing the treatment allocation were prepared prior to the trial. It is not explicitly stated however, if the allocation was concealed from investigators and participants. Blinding: outcome assessors were blinded to treatment allocation. Losses to follow‐up: one drop‐out reported, from the oral steroid group. Sample size calculation: not reported. Appropriate statistical analysis: insufficient data reported to determine if intention to treat analysis was performed.
Participants	28 participants included, appears 37 were screened for inclusion. Inclusion criteria: new, not previously treated episode of frozen shoulder syndrome or adhesive capsulitis; defined as painful limited, passive glenohumeral mobility, external rotation <30 degrees, abduction <90 degrees and in stage 2 or 3 of the disease process; stage 2: external rotation between 20 to 30 degrees and abduction between 60 degrees to 90 degrees; stage 3: external rotation less than or equal to 20 degrees and abduction less than or equal to 60 degrees; age 40 years or over. Exclusion criteria: contraindications to oral or intra‐articular injection of corticosteroids, insulin‐dependent diabetes mellitus, neurological disorders, cervical spondylosis, previous fracture or surgery or dislocation of the shoulder area, severe gastric complaints.
Interventions	Group 1 (15 participants): oral corticosteroid treatment, triamcinolone 4mg tablets, taken orally 3 times per day for 1 week, then two times per day for 1 week, then once per day for 1 week. Group 2 (13 participants): intra‐articular corticosteroid injection of 40mg triamcinolone acetonide using the posterior route Participants in both groups had physiotherapy from day 4 of 12 sessions of 20 minutes each, consisting of active exercise and passive joint mobilization. Participants could use ice or hot packs, but no other forms of treatment were allowed.
Outcomes	Outcomes assessed at week 1, 2 and 3. 1) 'Cure rate' of participants: defined as able to achieve 90% of normal passive glenohumeral range of motion for abduction and external rotation 2) Pain; assessed on a visual analogue scale; the authors do not report details of the scale. 3) Adverse events: number of cases of epigastric pain and number with injection site pain.
Notes	Cumulative proportion of 'cured' participants in each group after 1,2 and 3 weeks is reported in a figure. Only baseline pain VAS scores and p‐values resulting from comparison of the (unreported) VAS scores between the two groups for week 1 and week 2 were reported. Number cured was extracted from the figure by the review authors and we have contacted the authors for a description of the pain VAS score and the mean and standard deviation in pain scores at each follow‐up. The authors report that the cure rate was higher in the injection group at one, two and three weeks. The VAS score was significantly different (does not state in which direction) after one week of therapy (p=0.022), but not after two weeks (p=0.239). Three participants in the injection group reported pain at the injection site and 3 participants in the oral steroid group reported epigastric pain. No external funding acknowledgements listed.
Risk of bias
Bias	Authors' judgement	Support for judgement
Allocation concealment?	Unclear risk	B ‐ Unclear

Characteristics of excluded studies [ordered by study ID]

Study	Reason for exclusion
Lloyd‐Roberts 1959	Not a randomised controlled trial.
Melzer 1995	Not a randomised controlled trial.

Contributions of authors

RB was responsible for writing the manuscript, performing the quality assessment and data extraction, analysing the data and interpreting the results of both the initial and updated review. 
 SG was responsible for performing the searches, selecting trials and performing the quality assessment for the initial review, analysing and interpreting the results and contributing to writing both the initial and updated review. 
 JY was responsible for extracting and analysing the data and contributing to writing the manuscript for the updated review. 
 RJ was responsible for updating the search and selecting trials for the updated review, and extracting and analysing the data and contributing to writing the manuscript for the updated review

Sources of support

Internal sources

• Department of Epidemiology and Preventive Medicine, Monash University, Melbourne, Australia.

• Department of Clinical Epidemiology, Cabrini Hospital, Australia.

• R Buchbinder is supported by a NHMRC Practitioner Fellowship, Australia.

External sources

• Australasian Cochrane Centre, Australia.

• Australian Satellite of the Cochrane Musculoskeletal Group, Australia.

Declarations of interest

Two of the reviewers had a conflict of interest as authors of one of the studies included in this review (Buchbinder 2004). To avoid bias, the paper was sent to an independent reviewer to assess whether it met the inclusion criteria for this review.

Edited (no change to conclusions)